How should i set model version?

Every time the ml backend is turned on, a random version number will be generated on the model version side, and the data and label results will be separated, which is very annoying
It is useless to manually add the version number to the json of the prediction result, it will also automatically give you the random number it generates
I know this question has been asked before, but I really don't want to export and import the whole project just to set up a model version, it's too much trouble
So.... anyone know how to do it?